RC bridge management optimisation considering condition assessment uncertainties

Abstract

Decision-making in bridge management has changed considerably in the past two decades and owners are additionally considering what types of interventions to implement, but correct decisions still need certain input. In Estonia, like in many countries, bridge management is based on inventory records and condition information. The main emphasis of this investigation is on improving the regular condition assessment. More accurate nondestructive testing methods and optimised inspection scheduling are proposed, to reduce condition assessment uncertainties. A conversion matrix for translating additional assessment results to the rating scale of the current Estonian Transport Administration management system is introduced and uncertainties in the condition state are analysed probabilistically. In addition, stochastic degradation models based on existing information are investigated to help considering uncertainties as a part of the overall management process. What impact the adopting of quantitative assessment, rather than qualitative visual inspection, may have on the suggested interventions schedule is also analysed. The probabilistic characteristics of the condition profiles of the most common bridge elements are computed using Markov Chain Monte Carlo stochastic simulation. The optimisation of inspection scheduling is performed considering the uncertainty of the initial deterioration model. When a threshold value, defined by the owner, is reached, the model is updated with assessment data to maintain the level of uncertainty below that threshold. The results confirm that deviations in the degradation model and assessment results influence the bridge condition uncertainty. Likewise, times of both inspection and intervention are influenced, which will ultimately impact the overall management reliability and costs.